On Fri, Feb 22, 2002 at 07:46:27PM +1100, Herbert Xu wrote: > Branden Robinson <firstname.lastname@example.org> wrote: > > > Tell it to Ben. My package just expects the symbols to be defined. > > I'll let the kernel and libc people play the blame game when it comes to > > deciding which of them should do so. > > > # include <sys/ioctl.h> > > And why do you expect that to give you the definition of every single > ioctl in the Linux kernel? Because ioctl's need to be unique anyway, and if the ioctl you want isn't available on the hardware (or you otherwise misuse it), errno is set appropriately. > FWIW, the SIS ioctls are not exported depending on whether SISFB support > is turned on. In fact, they don't get exported at all, they live in > drivers/drm/drm.h. > > Getting them from glibc makes as much sense as getting them from KDE > because glibc has no relationship to the kernel that you're running. Hrm, yes, like how KDE wraps system calls like ioctl() for me, and glibc doesn't... Like I said, talk to Ben. I get shit on when my package Depends: on kernel-headers-*, I get shit on when it doesn't. > The recommended practice is to define the values in the programs that use > them, in this case, X. It makes even more sense in this case as the DRM > drivers are actually maintained by the XFree86 Project. I didn't hack up XFree86 to make it do things this way. Send me a patch or tell email@example.com what idiots they are. Directing your condescension and abuse at me is moronic. (And, in actual fact, the DRM modules are maintained by the Mesa guys; XFree86 periodically syncs up with them. But since you've made up your mind I won't try and confuse you with the facts.) -- G. Branden Robinson | When I die I want to go peacefully Debian GNU/Linux | in my sleep like my ol' Grand firstname.lastname@example.org | Dad...not screaming in terror like http://people.debian.org/~branden/ | his passengers.
Description: PGP signature